People on Starship - Sooner than You Expect
Вставка
- Опубліковано 20 бер 2022
- SpaceX seems to feel that Starship will fly people in the near future, but other people feel that the lack of an abort system means it can never fly people.
This video looks at the reliability of existing NASA launchers and explores how the reliability of Starship compares to those other vehicles.
@Eric_Gunnerson on Twitter
Triabolical_ on Reddit - Наука та технологія
Fun fact: 2 out of 3 Merlin fails weren't even Merlins fault, flexible shroud failed.
Raptor shrouds are solid.
Another fun fact: Falcon 9 failed a mission 2 times because someone thought it is a good idea to put helium COPV inside LOX tank.
All Starship COPV's are outside of propellant tanks.
And, of course, Starship *probably* doesn't have any helium COPV...
@@EagerSpace
It does, but it doesn’t use them much anymore, and there aren’t nearly as many on the booster because now the launch pad uses smaller individual QDs connected to the outer ring of Booster engines which feeds oxygen/methane gas to start them up. At some point though the COPVs could be completely eliminated, if they figure out a way to not need to start the spin-starting engines with helium.
I think you are overlooking the potential for engine failures and underperformance to be correlated events, plus things like increased risk in engine-out scenarios. This is particularly relevant during landing where even brief underperformance is instant death.
One of the key advantages of an escape capsule is that it is dissimilar redundancy, not likely to fail for the same reasons as, nor dependent on, other critical aspects of flight. One of the key advantages of reentry in a capsule is that the simplicity of the mechanism reduces the number of individual parts that can fail-if you have your heat shield and at least one parachute, you live. These components are as simple and as overbuilt as they could reasonably be.
That said, I agree when comparing to non-Dragon crew launch that there is a good argument to favor Starship. Rockets that don't launch much can never have real safety factors that are extremely hIgh.
I touched on that when I talked about systemic issues, but probably not enough. I had a section on fratricide and correlated issues, but decided to pull it out. The limited research I did suggested that these scenarios are really quite rare; even in development rockets the issues that crop up are largely engine-specific.
Dissimilar redundancy is good, though I would argue that parachutes are anything but simple; they fail fairly often and the failure scenarios can easily cascade. There may be another video there.
Its good to see reasonable discussion on risk rather than the "if it saves but one life" mentality.
Looking back at the real world shuttle fail rate 2 in 135 or 1 in 67.5? it seems kinda crazy. I think the other factor we must consider is crew compliment. 7 people KIA in one shuttle mission was the factor that made those particularly painful to stomach. How would the world react to a starship with 15 or even 20 or more people lost on a single mission.
Read Rand Simberg's "Safe is not an option" ebook. It's $5 or so on amazon.
If we want to have a real space exploration program, people are going to die doing it, and that needs to be understood. 15 people die every day in the US in on-the-job accidents and 95 people die every day in car accidents, but that is not newsworthy because we have accepted it as a society.
NASA is part of the problem here; they regularly use the word "safe" when it comes to space. It doesn't really apply.
We will see how things play out. I won't be holding me breath. God speed to the Polaris 3 crew.
I’m quite glad that people like you make videos like this so that people who become critics of Starship with normally quick responses have to educate themselves to create valid counter arguments.
The tenth digit of pi is a 3. Memorized pi to 300 digits as apart of a competition.
I hoped somebody would comment on that....
The tenth digit is 3, but the next digit a 5. So if you round up, it's a 4, but yes, 4 isn't strictly a digit of pi.
I can memorize pi to 3.00 digits!
So I will say, I think you overcalculated the odds of Starship safety here a couple of times (although maybe under-estimated it in one way too). So the first thing I'd like to point out is the 1/500 engine failure claim. Raptor is a significantly more complicated engine than Merlin and operating far closer to it's material limits. I would not be convinced it would be as reliable as Merlin, at least not initially.
The second thing I'd like to point out is a recalculation in your favor, that an abort of mission would be a loss of crew. Given that Starship has the ability to do powered landings, it maybe have some RTLS or ditching options to prevent a LOC.
The third thing I'd like to point out is that the current landing sequence seems to be a 3 engine flip 2 engine landing, and that a 1 engine flip is impossible (as per SN9). This would allow Starship to lose 1 engine on either the flip or landing, but if it lost 2 in either procedure it would be doomed. It could only survive losing 2 engines if it lost one in the flip and one in the landing.
And the fourth thing I'd like to point out is that SN11's failed landing was due to a propulsion explosion. Now sure, it was related to the engine burning a vital part of it's starting sequence beforehand. But if we're talking about rapidly reusing boosters with little to no engine inspection, and keeping the ascent and landing engines effectively separate on orbital Starships, there is a chance that the engine damages itself in a way that'll explode on an ascent rather than a descent, and I think brushing this problem off as rare when the Starship prototypes showed Raptor to be susceptible of this sort of failure isn't great.
Also, quick wikipedia googling. Seems like the risk of LOC on the ascent and descent phases of Commercial Crew capsules can't exceed 1/500. The overall mission is just lower from both combining a launch and landing as well as the risk of micrometeorites and debris impacts in orbit.
First thing, absolutely. I meant to put a disclaimer at the beginning that said that there's a lot of slop in the estimates and for some reason forgot.
Second. Yes, and that's a good future subject.
Third. Yes. I don't know that I engine flip is impossible and I don't know if they can do it aerodynamically or with hot gas thrusters. I *suspect* that there's more controllability there than we've seen, and I also suspect that they will be carrying higher fuel margins on crewed flight and will do a flip that is much higher.
Fourth. Yes. My argument is that there is every reason to suspect that Raptor will be just as reliable as Merlin or the RS-25 once it is developed, and that the experience with development engines is not significant. Or, to put it another way, both Raptor and Starship were at a low level of refinement in the tests that we've seen.
Last. Yes, 1 in 500 my understanding, though it doesn't make sense to me; if they are each 1 in 500, then in aggregate they can't be any better than 1 in 250, they would have to be quite a bit better. And yes, the hard part for commercial crew is apparently not ascent/descent but the on orbit stay; there is simply quite a lot of impact risk and it's really hard to mitigate. There was a lot of discussion during development about whether those risks made it impossible to meet the goal and what models would be used.
"The third thing I'd like to point out is that the current landing sequence seems to be a 3 engine flip 2 engine landing, and that a 1 engine flip is impossible (as per SN9)"
I wouldn't extrapolate that far from a single test flight.
A single Raptor 2 easily provides the necessary TWR, the harder question is wether or not a nominal landing profile can transition to a single engine profile in flight.
The problem with this failure mode analysis is that it treats engine failures as independent events, which they are not.
The engines are packed tightly together, with both systems and basic physical correlations in their failure modes.
There are numerous failure modes that could take out more than one engine.
So the system isn’t nearly as safe as this analysis suggests.
Whether it is “safe enough” is another question entirely, as you describe so well.
Yes.
The real analysis looks at all the failure modes you can come up with and tries to estimate how likely each of them is.
That's a very involved process and the only group skilled enough to do it are the engineers that designed the vehicle.
I talk about this a bit more in this video:
ua-cam.com/video/Gdi3lebIwWE/v-deo.html
Goddamn that pop quiz is hard. Definitely putting the shuttle in last. Saturn V is probably 4th. But man you could argue anything for the remaining 3. Soyuz has history, a lot of time to perfect the systems, and a perfectly serviceable escape system. Include the early launches and I think it's in 3rd or even 4th, but only look at modern Soyuz and that's a different story. I'd probably put it in third the more I think about it. Falcon 9 and SLS/Orion each have their own advantages and disadvantages; Falcon 9 has a full-launch-profile abort system, but it's liquid fuel based and on the sides of the capsule, which given that explosion in testing (even though that root cause was found and corrected) does pose a bit more of a risk than a solid-fuel launch tower like Orion. However, SLS/Orion has SRBs, and going to the moon also doesn't offer the safety benefits of LEO. So maybe Falcon reigns supreme here?
Guess I didn't do well here, although I guess factoring in the lunar missions for Apollo and SLS changes things by a large margin rather than weighing them more heavily by ascent safety and partially brushing off lunar missions.
Saturn V on an easy mission is arguably better than SLS on the same mission. .
Soyuz hasn't killed anybody in *50* years, though the current quality has been going downhill.
SLS is - in the words of the ASAP - disappointing. NASA has been very cagey about LOC studies for SLS - in contrast with Constellation, when they had some in depth ones. Their *goal* is 1 in 300 for ascent and entry (so, less than commercial crew) and 1 in 150 for the cislunar mission. I haven't been able to find the details, but it's not especially clear to me why a cislunar mission is riskier than a 6 month capsule stay at ISS; there is a *lot* of debris risk there. But yes, SLS does get a bit penalized by doing something useful rather than hanging around in orbit like shuttle.
WRT Dragon's escape system, I agree that more hypergolics makes it more dangerous - though all capsule use them - but remember that it doesn't have the issue where the escape rocket fails to jettison.
very interesting video, but many many assumptions. starship is not planned to have 9 engines. elon only mentioned the possibility. also, the raptor engine is a completely seperate engine from merlin and still in the prototyping phase. trying to estimate a chance of failure per engine for the final design is pretty much impossible. but knowing spacex, they will propably go for something similar if not better than what merlin achieved
Yes, lots of assumptions.
Musk said "The ship is just begging for three more vacuum engines to be added. Tanks will stretch for more propellant load." That's been widely interpreted as them going to 9 engines, at least for some variants. I think it's highly likely for the crew variant as 6 engines means thrust/weight is less than 1 (even with raptor 2, I think).
I don't think it's unreasonable to think that Raptor will be as reliable as Merlin or the RS-25, and that's why I chose 1 in 500. I actually expect it to be better than 1 in 1000 pretty easily.
@@EagerSpace yeah, that makes sense
Here are my quiz answers answers:
Soyuz, Falcon 9 + Dragon, SLS + Orion, STS Shuttle, Saturn V
Oh, wow! I had expected way better from SLS and especially from Soyuz.
Besides that, because of the iterative approach SpaceX pursues with its rockets I think there's a way to drive the risk of a crew loss to virtually zero even relatively early in the program:
Falcon 9 has perfected the reliability of its recovery by flying a lot and occasionally losing the rocket. This way they can find niche and unexpected failure modes and fix them for good. It's clear that SpaceX is approaching Starship development from a similar angle. Now, if the majority of Starship missions will be uncrewed, the chance of such failure happening on a crewed one becomes substantially lower than the chance of a critical failure happening overall. And afterwards these probabilities get reduced even more (because the underlying design flaw gets fixed).
can you do a starship abort modes? ( similar to the shuttle)
Yes. That is in fact the first video I started on, and that one spun off the shuttle abort modes video, the safety at NASA video, and this video.
But the abort mode one is coming.
Really excellent breakdown of risk, there seem to be a lot of negative but unsubstantiated opinions around, thanks for helping to clear the FUD.
There appears to be a disparity between what NewSpace is expected to Demonstrate and the Pre-approval of SLS, Orion.
Many also seem to not grasp that StarShip/HLS can achieve many mission targets prior to crewed launches & landings.
The hardest problem people are having is separating the issues they've seen with the test vehicle from the likely performance of the production vehicle. There's also this unstated belief that parachutes are perfect when in fact they have quite a few issues.
If you haven't watched my other video on safety practices at NASA, you might enjoy it. The short form is that NASA has been very good at presuming that they are the best at safety while mostly ignoring the best practices.
I don't find myself agreeing with all of your assessments but you do a good job speaking your point.
I don't think *I* agree with all of my assessments; there's a lot of shaky data going into them.
Regarding the risk of breakup on ascent due to engine failure, I think you undercount them. Challenger was effectively that scenario, for instance, and Starship itself broke up due to engine failure (albeit on landing) on one of the high-altitude test flights.
My goal was to look at failures of liquid fuel engines as that is what Starship uses.
The SRBs on Challenger failed because they had a design that was not redundant in their usage (the casings flexed because of the weird load that shuttle put on them), because they were very cold and the actual specification for temperature of shuttle was not defined, and because there was a very strong wind shear on ascent that made the flexing worse.
WRT Starship having engine issues on test flights, failures during engine development are not indicative to problems with production engines. The RS-25 failed all the time in development and the F-1 was notorious for instability, but they were quite reliable in production use.
I know this is basic math, but you put it across so well that it's worth it.
Maybe they should have uhhhh, ejection seats (: , heard they get extra performance in a pure oxygen environment!
1 in 500 failure rate for raptor 2 maybe a little optimistic initially given novel architecture
Colombia was not the tiles failing. It was a breach in the RCC wing edge. Starship does not have these. No reason discussing tiles. Tiles have never been the cause of a lost vehicle.
The probabilities of engines failing do not seem like independent probabilities to me. If 1 or 2 engines fail then we all know that the risk of a 3rd or 4th engine failing is significantly higher. You can also argue that what is often considered an engine failure on a rocket with one or a few engines, is quite often a (limited) system failure rather than an independent engine failure.
The question is whether the risk of cascading failures is bigger than the benefit you get with redundancy, and I don't think we have sufficient data to know.
We do know that Merlin has been ridiculously reliable on both Falcon 9 and Falcon Heavy, and I don't think there's any reason to expect that raptor is going to be significantly worse than Merlin.
@@EagerSpace We have indeed already seen two launches where all Raptor engines worked correctly, so hopefully Starship will have good reliability in the future too.
Can starship actually land with one engine operationally and not just theoretically?
The Starship landing burn appears to start with all three engines burning, presumably in order to reduce gravity losses, and then cut down to a single engine for final approach.
Most of Starship's vertical velocity is actually cut by that initial high thrust three engine burn, and it would take much longer and more fuel to do it with fewer engines.
I actually looked into this as a a result of wondering why they chose to go with three raptors despite having sufficient TWR to land on only one, the answer -besides potentially redundancy- is considerable fuel savings from reduced gravity losses, probably both on ascent and descent.
If the engine fails on ascent, the reduced thrust would result in more gravity losses to achieve target orbit and a larger fuel requirement for landing.
In combination this would probably necessitate an abort of the mission, but it could well be survivable.
If an engine fails to re-light on a regular landing approach however, it might not be survivable. The increased time to flip and the lower acceleration mean that unless there is sufficient margin allowed, starship will not decelerate in time, with two engine failures (obviously unlikely) it's much harder still.
Maybe allowing for this margin could be a contingency for human flights.
Didn't mean to put a strike through the redundancy comment.
Initial SS landings used 2 engines 3 was adopted to allow polling of best engines, for landing, and switching off what is not required
Cool stuff, but I highly doubt Raptor 2 will have "already doesn't" have a failure rate of barely 1/500, the engine is much more complex and burns far hotter then the Merlin engine, according to Elon himself they are still having issues with the throat trying to melt itself even with cryogenic fuel flowing between it. Also Starship doesn't need all three engines to fail to not be able to land/catch, for example if an engine or two fail in the middle of the flip/reorientation the vehicle it wont be able to cancel out its flip velocity and over-flip, you can see this with SN9.
WRT failure rates, the RS-25 is a *far* more complex engine and the hydrogen turbopump is particularly highly stressed, but NASA achieved 1 in 400 with an engine designed in the late 1960s. We are simply seeing an engine that isn't fully developed yet, and the issues that show up in development are not indicative of issues when deployed.
I went back and watched SN9 a few times. I don't see anything that precludes success on a single engine, though I would agree that failure of an engine during the operation might be problematic.
SN9 isn't really a great example. After that test flight they switched to firing all 3 engines then down selecting to the single highest performing engine. And yes, Starship landings are doable with a single engine, see SN5 and SN6.
Great video as always. The first ten digits of Pi are kinda easy to remember: 14 15 92 654 a bunch of sequences plus you oughta know or know of someone who was born in 92.
Colombia was not the tiles failing. It was a breach in the RCC wing edge. Starship does not have these. No reason discussing tiles. Tiles have never been the cause of a lost vehicle.